Forecasting Time Series

Abstract

This chapter includes two problems for forecasting of a time series using past data points. It is argued that the past data points used for forecasting of the future data points should be strongly correlated with each other. It is illustrated that the strongly correlated past data points can be identified from the autocorrelation function of the time series. It is further illustrated that a powerful forecasting procedure for the time series can be a recursive technique. Its application is demonstrated using, as examples, annual patient volume forecasting, as well as forecasting of the seasonal variation of the hemoglobin A1C level.